Yousef S.

I am Youssef from Egypt. I am 20years old.

I build computer vision systems that work beyond the notebook: image classification, object detection, and OCR pipelines, properly benchmarked and demo-ready.

What I’ve delivered

• Image classification: benchmarked DINOv2, EfficientNet, ResNet, and YOLO on a real dataset. The best model reached 95%+ macro-F1 on a controlled validation set, with Grad-CAM explainability and a live demo on Hugging Face Spaces.

• Detection + OCR: contributed to an industrial YOLOv8 detection and OCR pipeline. Deskewing and cascade fallback raised accuracy from ~54% to ~93% through systematic error analysis.

How I can help

• Image classification, object detection, OCR

• Model comparison and fine-tuning

• Data cleaning and ML modeling

• AI feasibility check: custom model, ready-made API, or no AI at all, followed by a scoped proof of concept before you commit a full budget

How I work

Define the problem → compare candidate approaches → pick one with a clear reason → build a working PoC → deliver documented code with a demo.

Stack: PyTorch, Hugging Face, YOLOv8, OpenCV, Scikit-learn, FastAPI, Docker, CUDA.

I currently work as a research assistant on applied AI systems, so what you get is benchmarked, deployed work, not tutorial-level projects. Send me your idea with a few sample images or data, and I’ll tell you honestly whether it needs a custom model.